<<

FLIGHT SYSTEM ARCHITECTURE OF THE SORATO 1 *John Walker 1ispace, 3-1-6 Azabudai, Minato-ku, Tokyo, , E-mail: [email protected]

XPRIZE (GLXP) as “Team .” Although ABSTRACT GLXP and Team Hakuto have ended without a ispace, a Japan-based private lunar exploration winner and without a lunar mission, Sorato is the company has developed lunar rovers over the baseline rover technology for ispace. course of several iterations, originally for its Lunar XPRIZE mission called “Team This paper presents the final architecture of the Hakuto”. A flight model rover called “Sorato” was 3.8kg Sorato rover, results of qualification and developed and qualified for the mission. briefly shows the conceptual design of the next ispace rover, carried to the lunar surface by its Sorato is a 4-wheel, skid steer rover with passive own lunar landers. ispace’s first orbiter mission is suspension and a minimum of actuators. The planned for 2019/2020 and the first landing/roving rover has two redundant controllers, inertial mission is planned for 2020/2021. measurement units, four cameras, and one time-of- flight camera for detecting obstacles and making 2 HISTORY 3D images. The rover also carries a deployable The conceptual design and prototyping for a antenna and has payload capacity. GLXP rover stretches back to 2010 at the Space Robotics Lab (SRL) of Tohoku University1 as Sorato is smaller than any lunar or martian rover, shown in Figure 1. at 3.8 kilograms. This mass was achieved by reducing the power consumption as much as possible, thereby reducing the surface area required for both radiators and solar cells, minimizing mass and by a novel thermal design.

The rover makes use of magnesium alloy, carbon fibre structure and low-thermal conductivity plastics, to thermally isolate the solar panels and lunar surface from the electronics. The top of the rover is concave which acts as a heater in the lunar morning and a radiator at lunar noontime.

Many parts in Sorato are Commercial-Off-The- Shelf (COTS), qualified through extensive Figure 1: The “Moonraker” 10kg four-wheel screening, especially radiation screening. The prototype rover, with the “Tetris” two-wheel entire system was designed, manufactured, and rover (inset) qualified at the system level. COTS components allow rapid development and higher performance ispace has since developed both two- and four- compared to traditional space products. wheel rovers, for independent operation as well as for cooperative operation to explore a lunar skylight. The rovers are described in Table 1. The 1 INTRODUCTION power consumptions shown are for 10% “duty ispace is a lunar exploration company based in cycle” driving unless otherwise stated. That is, Tokyo, Japan. It intends to travel to the lunar 10% of time is spent driving, and 90% of time surface, find and utilize resources, particularly making decisions. This number is chosen based on water- which is thought to exist at the lunar ispace’s field test experience for operation at sub- polar regions. ispace plans to do this in a cost- 100kbps bandwidth. effective and profitable way through its expertise in miniaturization of spacecraft and lunar rovers, All rovers include ispace’s qualified motor beyond the lower bounds established by space controllers and Inertial Measurement Units agencies. (IMUs). All of the four-wheel rovers use the same basic principle of 4-wheel skid steering, with a ispace created a four-wheeled lunar rover called simple passive differential gear system to keep all Sorato as part of its entry into the Google Lunar four wheels on the ground. Table 1: History of ispace rovers since 2013. Name Description “Pre- Description: 4-wheel rover Flight Camera: 5MP omni-directional camera + mirror Model” Radio: 900 Mhz proprietary + 2.4 Ghz wi-fi (1 (PFM) antenna for each) Controller: FPGA space-ready main controller + ARM imaging controller Mobility: 4 in-wheel motor Solar Power: none Battery: space ready Li-ion, 77Wh Mass: 7.0kg Power: 18W

Description: 2-wheel rover (tethered) Figure 3: The “FM1” Sorato four-wheel rover Camera: 5MP Radio: 900 Mhz proprietary + 2.4 Ghz wi-fi (1 In PFM, the omni-directional camera system was antenna for each) Controller: ARM controller usable but its height required a tall support Mobility: 2 in-wheel motor structure to support the rover during launch, which Solar Power: none added significant mass. It was replaced by 4 Battery: space ready Li-ion, 38Wh cameras with overlapping fields of view in PFM2. Mass: 1.5kg Power: 10W PFM had a heterogeneous architecture, with a “Pre- Description: 4-wheel rover high-reliability FPGA-based controller plus a Flight Camera: 4x 5MP COTS ARM controller to meet the imaging Model Radio: 900 Mhz proprietary + 2.4 Ghz wi-fi (1 2” antenna for each) requirements. The ARM controllers passed (PFM2) Controller: 2x ARM controller (redundant) radiation screening so PFM2 was updated to a Mobility: 4 in-wheel motor redundant ARM controller architecture. This Solar Power: None general concept was kept until FM2. Battery: space ready Li-ion, 77Wh Mass: 7.0kg Power: 15W The FM1 rover was designed for the thermal “Sorato Description: 4-wheel rover environment at 45N and was qualified at the Flight Camera: 4x 5MP component and system level2, especially by Model Radio: 2.4 Ghz proprietary 1” Controller: 2x ARM controller (redundant) mobility, thermal vacuum, and vibration testing. (FM1) Mobility: 4 in-wheel motor Testing was used to validate analysis models used Solar Power: 18W @ 45N for FM1, and allow three major, related changes to Battery: space ready Li-ion, 77Wh Mass: 5.0kg be made for the FM2 rover: power reduction, Power: 13W thermal design, and mass reduction. “Sorato Description: 4-wheel rover Flight Camera: 4x 5MP Partway through the FM2 design, the landing Model Radio: 2.4 Ghz proprietary 2” Controller: 2x ARM controller (redundant) partner was changed and the landing site was (FM2) Mobility: 4 in-wheel motor changed from 45N to 28N. This hotter location Solar Power: 11 W @ 28N required improved thermal design by less heating Battery: space ready Li-ion, 38Wh Mass: 3.8kg and more cooling. Less heating was accomplished Power: 13W (at 20% duty cycle) by lower power consumption, primarily by using more efficient motors. Less power consumption allowed smaller solar panels and batteries. Smaller solar panels allowed a shorter rover. More efficient motors, smaller batteries, and a shorter rover all contributed to a lower mass, along with change to a monocoque structure. 3 SORATO FM2 ROVER 3.1 Requirements The requirements for the rover came from four main inputs: 1) The expertise and business Figure 2: The “PFM” two- and four-wheel rovers planning of SRL and ispace, 2) The GLXP requirements, 3) The specific environmental Since the PFM rover, all components have been requirements of the mission and landing site and qualified through radiation (total dose, single 4) Constraints of the landing partner. The landing event), thermal vacuum and vibration testing, and site at 28N was selected by the landing partner as system environmental testing. Progress was made a relatively safe site with few rocky outcrops and by reducing the power consumption with each with relatively good available terrain data. Table 2 iteration, thereby decreasing required power, solar gives a summary of the requirements. All of these panel size, and total rover mass. In each phase, were met through design or testing. screening and qualification testing was performed. Table 2: Top level requirements for Sorato Source Category Requirement Total mass of the rover is 3.8kg (without the GLXP Imaging NRT Video “interface” to the lander structure. The maximum HD Video High res panorama speed is approximately 0.1m/s and the power PR GLXP logo shown consumption is 4.5W at idle (including Looks cool communication) and an average of 21.5W while Mobility Travel 500m total moving across varied terrain. Travel 250m from lander ispace Mobility 22 degree slope climb The rover is designed to be dropped from the deck Overcome 10cm rocks Autonomous stop of the lander to the lunar surface. The component 0.1m/s speed which holds and drops the rover is called the Budget Mass < 4kg interface, shown below. It has 3 main functions: PR Sponsor logos deployment using a COTS actuator, thermal Power 10% driving time isolation of the lander to the rover, electrical Landing Thermal Surface temperature to 100C site Power Sun angle to 28 degree interface for heaters inside the rover (used during Radiation Up to 80MeV cruise). Landing Radiation Up to 4krad total dose Partner Structure > 60 Hz eigenfrequency 3.3 System Architecture 20G quasi-static Mission time (proprietary) Aside from connecting the subsystems, the Thermal Isolate from lander defining feature of the architecture is the two Comm 2.4 Ghz, 1W radio redundant ARM controllers located on one ispace- 125 bps uplink (33% of designed board. time) 118kbps downlink (50% of time) CCSDS, Proprietary ground station

The GLXP imaging requirements are detailed but essentially require a narrow field of view, with 1- 2fps “Near-Realtime” (NRT) 640x480 video for operation and 10fps 1280x720p HD video sent to Earth (non-realtime)3. Based on the above requirements, the mobility, thermal, and structure detailed designs were created.

3.2 System

Figure 6: Sorato Rover Architecture

3.4 Mobility The four-wheel skid steer concept was kept as it has been successful for ispace. The mobility system comprises an ispace motor controller, 4 COTS (but customized) brushless DC gearmotors inside each of four wheels, and ground clearance and wheel size maximized for the available space on the lander. The wheels are mass-optimized based on previous research at SRL which concludes approximately 15 grousers with 15mm length is an optimum point for travel efficiency vs 4 performance . The mobility system can generally Figure 4: Sorato Rover, antenna deployed traverse over 10 cm obstacles or higher and remain stable at slopes approaching 30 degrees. 3.5 Thermal Design The interface is designed to thermally isolate the rover from the lander. This is accomplished by MLI placed inside and outside the interface and ensuring the contact points of the rover near the lander are as small as possible. During cruise, the Figure 5: Dimensions of Sorato, antenna stowed lander provides power to several heaters inside the rover, particularly near the battery. interface is constructed from CFRP with Nomex honeycomb, and is approximately 20mm thick in During the surface phase of the mission, the total. surface of the ranges from about -40 to 100C at lunar noontime. Therefore, the bottom of The rover is constructed mainly from CFRP with the rover is insulated by MLI (inside and outside Nomex honeycomb, and is approximately 3.5mm of the monocoque). The bottom and sides of the thick. Note that the requirements for the vibration rover become hot due to the solar panels, therefore environment are largely met by the interface, all electronics are mounted on top of the rover, while the rover mass is minimized. Other directly to magnesium radiators with a silver- components are generally machined from low teflon thermal control surface. The top of the rover density, high thermal conductivity magnesium or is essentially magnesium, and is thermally isolated titanium where strength was required. from the rest of the rover by Ultem plastic. The wheels are made from the same plastic, to thermally isolate the rover from the surface. The top of the rover can maintain a temperature 50C lower than the bottom of the rover.

Figure 9: Sorato inside the interface, shown without MLI for clarity

Figure 7: Electronics are located on the top of the rover

The top of the rover has a concave shape comprising near-vertical sections with high absorptivity and horizontal sections with low absorptivity and high emissivity. By this design, Figure 10: Sorato “bottom” showing monocoque the top of the rover acts partially as a heater when construction, mid-integration it has a high view-factor to the sun while it is low in the sky, and acts primarily as a radiator when 3.7 Imaging the sun angle is higher. Four cameras are connected by the cameras’ native interfaces to the Image Processing Unit (IPU) of the controllers. This is key to reduction in power consumption and is based on smartphone architecture. A conventional robotics camera has a local controller. By using a “native” interface connected to the IPUs, the local controller and most of the data handling is avoided, saving power. The IPU obtains frames from the camera and hardware compresses them as the first step. Raw frames do not need to be handled by the

Figure 8: Detail of the heater function of the CPU. Unlike typical space applications, Sorato radiator uses H.264 movie compression, not still images

3.6 Structure A Time of Flight (TOF) camera is also included Aside from the thermal design, the interface is on the front of the rover. It is required to detect designed to make the rover as stiff as possible obstacles which may be undetected by the operator and may be dangerous for the rover to traverse when connected to the lander. To do this, there are 12 points which constrain the rover, plus one (i.e. less than the 10cm height the mobility system actuation point in the middle of the rover. The can easily handle). The TOF camera is a COTS design customized for ispace with a laser light TOF are both displayed on an overhead 2D map, source more suited to the lunar lighting and overlaid on the camera images. environment. It functions to 2-3m and can create a point cloud for the operator to download and/or generate a local map of obstacles and perform emergency stopping function. The overlapping combination of this function plus mobility over obstacles is key to rover success.

3.8 Communication The communication system was heavily constrained to the landing partner’s lander design and ground station design. The most difficult constraints include a maximum downlink speed of 118kbps, time multiplexed to 50% at the sub- Figure 11: Sorato pilot screen (1 of 2 screens) second level and a maximum uplink speed of 125bps, time multiplexed to 33% at the second level.

The proprietary 2.4 Ghz radio was made from COTS components, but required extensive testing and design updates to reach the required performance. The specific details of packetization for a CCSDS compliant system and time multiplexing are beyond the scope of this summary report.

Figure 12: Sorato co-pilot screen (1 of 2 screens) Meeting the distance requirements of GLXP with margin and varied terrain was done by selecting an Telecommands are uploaded via a ground station optimum antenna and placing it as high as network and lander and telemetry/images are practicable on top of the rover. The top of the downloaded via the lander and a ground station antenna is 700mm from the ground. This is network. The detailed communication architecture achieved by developing a titanium spring is out of the scope of this report. mechanism and using a COTS actuator to deploy it on the lunar surface. 4 COMPONENT SCREENING The specific components for Sorato were chosen based on past experience (updated from older rovers), primarily as COTS at the small component level (for example, CPU, memory, and camera sensor). The PCBs for all subsystems were designed by ispace and manufactured in Japan with the exception of the power distribution unit, which was procured from a cubesat supplier.

The following types of component level qualification was carried out: 1) Vibration testing to 25Grms, 2) Thermal vacuum test for maximum Figure 10: Sorato, antenna stowed and minimum expected temperatures, 3) Thermal cycle test (150 times, -40-80C), 4) X-ray 3.9 Operation inspection (for COTS PCBs with ball-grid array The rover is designed to be operated primarily by components), 5) Single event effect (to 80MeV), one pilot and one co-pilot. Additionally, one 6) Total Ionizing Dose test (to 20 krad). mission manager handles mission planning and one or two system or support engineers are This testing was done many times, when a available to support mission planning based on candidate for a component was selected and when operational constraints, temperatures, power the designed was finalized. Historically, requirements, etc. The screens offer an interface to approximately 70% of selected components passed send commands for driving, check all telemetry screening. Over time, the following For the Sorato and alarm conditions, receive images, video, FM2, 100% of selected components and 100% of obstacle detection information and download final designs passed. additional data such as point clouds when required for additional analysis. Obstacles detected by the 5 SYSTEM QUALIFICATION orientation, etc. The updated models confirmed The qualification strategy was based on producing confidence in the design for the mission. 3 rovers: 1) Flight item, 2) Flight spare: identical to above, 3) Field test rover: Nearly identical to Table 4: Thermal-Vacuum test levels for Sorato the above, but some added components for field Name Description testing including RTK-GPS, debug connectors and Cold With rover inside lander interface and an additional battery to simulate solar charging on case simulated lander. Chamber Shroud kept at minimum temperature, “Lander” kept at - Earth. Solar cells and thermal gear was not 40C and heaters enabled installed. Hot With rover and simulated lunar surface. case Surface kept at 100C and IR lamps used to 6.1 Environmental Testing simulate maximum solar cell input and System level environmental testing of Sorato was solar flux input to rover structure and performed. The test levels for the mission are radiators proprietary, but the following levels were used for thermal vacuum and vibration testing at the The following image summarizes the system level: environmental testing program.

Table 3: Vibration testing levels for Sorato Name Description LL Low level (0.5G or 0.5 Grms), for evaluating eigenfrequency response before and after higher level tests MSL Minimum Specification Level, for verifying that a configuration change does not have a workmanship issue AT Acceptance Test level, approximately the expected level in flight. PFT Proto-flight level, above the expected level in flight

Each vibration testing level includes a specification for random vibration testing and sine-sweep vibration testing and differing levels for each of 3 axes. The full vibration tested program was designed with the launch provider to Image 13: Test Philosophy of the 3 rovers test to the local environment of the rover in the lander system, the 20G quasi-static requirement, Both the flight item and flight spare were testing and to verify the eigenfrequencies against the in a sequence of thermal vacuum and vibration to predicted values and requirement. simulate the launch sequence. After successful deployment testing, the flight item was re- The protoflight level was used rather than a assembled and vibration tested to MSL to ensure “Qualification Test” or “QT” level used on some no change in stiffness, and then shipped for programs. Protoflight level is somewhat higher integration with the lander. All vibration testing than AT, but lower than QT. This gives was successful, with no damage indicated and all confidence to all stakeholders that one spacecraft rover functions working before and after testing. was tested beyond the predicted levels (there is Deployment was also successful. safety margin) but has a lower level of risk to reduce the spacecraft lifespan or reliability. This is 6 FIELD TESTING important for a minimum mass design and Field testing is useful for evaluating mission important for a “newspace” project to reduce the planning, user interfaces, operator training. It is number of spacecraft required. Minimum partially done in a quantitative way (for example eigenfrequency of 58 Hz was determined. formally checking all rover functions within the entire system), and partially done in a qualitative Thermal-vacuum testing was performed with the way. interface and simulated lander for the “coldest case”, which is during cruise. It was performed 6.1 Quantitative Field Testing with the rover and simulated lunar surface for the Field testing, defined as physical system level “hottest case” of lunar noontime. Predicted testing in an analogue environment was performed temperatures were within 10C, and all rover in 3 locations: the SRL facility’s sandbox, the functions worked before, during and after testing, indoor field test facility at JAXA’s facility in verifying the design. Thermal-vacuum testing was Sagamihara, Japan, and a natural sand dune site at also used to calibrate the thermal analysis models Tottori, Japan. Table 4 gives a summary (at and predict results at other times of day, rover subsystem and general topic level of the quantitative field testing results):

Table 4: Quantitative field testing results Location Test Result SRL Slope to 22deg Pass (stable) Max slope 25deg (slip ratio=0.6) Saga- Imaging to Pass mihara specification 10cm obstacle Pass Characterize obstacle Identify dangerous performance rock sizes Figure 16: Usability testing during field testing at TOF obstacle detect >5cm Tottori, Japan, showing all 4 screens Tottori Operation at 250m Pass

500m operation Pass All telecommand Pass All telemetry Pass Imaging for mission Pass planning TOF obstacle detect Pass during operation

Figure 14: Maximum slope testing at SRL Figure 17: UI showing obstacle detection overlay on image and position on 2D map 6.2 Qualitative Field Testing During field testing, several other findings were made. Most importantly, the image quality, and 7 CONCLUSION especially NRT quality which met specific All requirements indicated in Table 2 were met, as requirements, was better than expected. Only shown in Table 5. Green items were qualified by 48kbps are available for NRT, but the images design and approval, and blue items were qualified obtained at 1-2fps were good enough for operators by design and testing. to run the rover for 20% of the time (double the design duty cycle). 1-2 cm variations in terrain and Table 5: Top level requirements for Sorato obstacles could be discerned. Source Category Requirement GLXP Imaging NRT Video HD Video High res panorama PR GLXP logo shown Looks cool Mobility Travel 500m total Travel 250m from lander ispace Mobility 22deg slope climb Overcome 10cm rocks Autonomous stop 0.1m/s speed Budget Mass < 4kg PR Sponsor logos Power 10% driving time (20% achieved) Landing Thermal Surface temperature to site 100C Power Sun angle to 2deg Figure 15: Frame obtained by the NRT video Radiation Up to 80MeV system at 2fps, clearly showing terrain Landing Radiation Up to 4krad total dose Usability testing was performed while operation of Partner Structure > 60 Hz eigenfrequency* the rover was underway with untrained operators, 20G quasi-static who could successfully determine the operating Mission time (proprietary) Thermal Isolate from lander procedures and constraints within several minutes. Comm 2.4 Ghz, 1W radio 125 bps uplink (33% of time) mission. International Astronautical Congress, 118kbps downlink (50% of Adelaide, Australia. time) CCSDS, Proprietary ground [3] GLXP rules, available at station https://lunar.xprize.org/about/guidelines * 58 Hz was achieved, which was deemed [4] Sutoh M (2013) Traveling Performance acceptable by negotiation with the landing partner. Analysis of Lunar/Planetary Robots on Loose Soil. Doctoral Thesis, Graduate School of The result of development of an easy-to-use UI Engineering, Tohoku University, Sendai Japan. and good results of image quality allowed operation at 20% duty cycle, as opposed to the 10% requirement.

7.1 Lessons Learned The Sorato program was successful, largely due to the opportunity to develop several iterations and improve in a step-by-step manner, as shown by qualification results. Outside of those results, there were “lessons learned” about COTS. 1) COTS components should be on the shelf. COTS electronics components have many advantages over traditional space components, but they require additional qualification, and often are discontinued or go out of stock. To use COTS components, one must be prepared to both update and re-qualify subsystems as technology progresses, and stock significant quantities to account for future programs and development. 2) One of the best advantages of COTS components is accessible software support. This should be one of the major selection criteria.

7.2 Future ispace Missions and Rovers The ispace business plan calls for a lunar orbiter mission in 2019-2010 and a lunar lander/rover mission in 2010-2011 followed by travel to the lunar surface once per month. This requires much better standardization of rover technologies and the ability to integrate payloads for customers quickly. As ispace plans to travel to polar regions to find and utilize resources (especially water), the first several missions are designed to build up the requisite technology for mobility and cold temperature roving. Therefore, the following requirements will be considered for the next rover: mobility in dangerous or vertical terrain (using the tethered system), standardization of mechanical, electrical, and data payload interfaces and operation in radio and sunlight denied areas. Over the course of several missions, extreme temperature, resource sensing and resource utilization will be added. References [1] Yoshida K, Britton N, Walker J (2014) Development and Field Testing of Moonraker: a Four-Wheel Rover in Minimum Design. In: proceedings of 14th International Symposium on Artificial Intelligence, Robotics and Automation in Space (i-SAIRAS), Montreal, Canada. [2] Walker J (2017) Final Configuration of the ispace Hakuto Rover for a Google Lunar XPRIZE